Jump to:

Philip Morris

Assessment of the Association Between Passive Smoking and Lung Cancer A Dissertation Presented to the Faculty of the Graduate School of Yale University in Candidacy for the Degree of Doctor of Philosophy

Date: 1987
Length: 101 pages
2023512852-2023512952
Jump To Images
snapshot_pm 2023512852-2023512952

Fields

Author
Varela, L.R.
Type
SCRT, REPORT, SCIENTIFIC
ABST, ABSTRACT
BIBL, BIBLIOGRAPHY
CHAR, CHART, GRAPH, TABLE, MAPS
Area
SCIENTIFIC AFFAIRS/BLACK LATERAL OLD S&T
Document File
2023512516/2023513116/Ets: Lung Cancer Volume I 930900
Litigation
Okag/Privilege Withdrawn
Okag/Produced
Characteristic
EXTR, EXTRA
ILLE, ILLEGIBLE
Site
R529
Named Organization
Ny State Dept of Health
Ny State Dept of Motor Vehicles
Ny State Tumor Registry
Sloan Kettering Cancer Center
Umi Dissertation Information Service
Who, World Health Org
Wk Kellog Foundation
American Cancer Society
British Medical Journal
Congress
Author (Organization)
Bell + Howell Information
Umi Dissertation Information Service
Yale Univ
Named Person
Akiba
Auerbach
Brunemann
Chen
Cornfield
Correa, P.
Dalagar
Enstrom
Friedman
Froeb
Garfinkel
Hirayama
Holford
Janerich, D.T.
Joubert
Kabat
Li
Mcnemar
Sandler
Trichopoulos
Vincent
Wald
White
Wynder
Yesner
Master ID
2023512517/3115

Related Documents:
Date Loaded
24 May 1999
UCSF Legacy ID
mmc02a00

Document Images

Text Control

Highlight Text:

OCR Text Alignment:

Image Control

Image Rotation:

Image Size:

Page 11: mmc02a00
ix 'fable No 21 . ........................................ UtOWR1SMM lSfLM1TF3. DL[rt3 RATIOS A40 CQR'IIA7iM LDGZS Fai A DiFFERFN3?J1L Q' DO'QSIPE DF 20_ UlIITS IN Y4E II4= OP E7POARE 'N 9DQAL PASSIVE SMOIWC. 121 'ifble i+b 22 . ................................. .... 126 ADJC6TE9 2grDa'>ZS. mD6 NATID6 AND CCWMEMM La415 Edt A DnYTAF71CS11L W l7XPOSittS Q 20 tRdT3 IN 'IHE II[t7t DF E7o'OSIItE 20 9QC7AL PASSNG 84'pQtC. LIST Or FIQLItFS PSQdtE Figure No 1. C1DNGE IId Ii1PG E7wCER IOG cODS WTit! NtlMBER DF QGXRFSTfs PfR D/lY SRUI® BY '1IM SPOUSE. EDR ALL SLMJFZ'P5. SfLF-ftFSRTDFMS. AtO SIRROa'!E RLS'PQrDEMSf AtD [DG CLDS EUR DDTICREIIf CJ1'i7EQtffS OF DG'06(RE. E'iytaeW2. CHD1tt= IN W= vuCEt ILG 0[06 UT3l1 lRUgER 0r CIGV03'= PFR DpY STO!® BY 'IYE S_ POLZE. E7t :.12. StJBJEZ"15. NES/tR 3T]DFS A!O fVll4R 9MOPMMt AND UG OC[1B EOR DZMSE?lC CJ1'lEGOtIFS aF EXPOSURE. Figure No 3. QIAPIEr IN it1NE WtcE7t LOG a006 MT1fF KPBFR cr CIC7kFtEiRES PER Dd1Y St7!® 4Y '1!E SFUWSE. FOR AIL SU9JfLT5. FIT111= MO FSRItS: AND IOG 0[XLS tt3R DIFFERtNi G1ECQtIlS OF E]XP051RE. Fiqure No 4. C3UMR= IN LUNC C7YCEP LDG 4006 r1TlH N!ffit CF CIGVLtTh.S PEP. DAY SlOKFD BY 'i1M SAOl15E. FOR AL*. SUBTF7CTS. IpIDF1UUID/s111i1. CZ1 . AMi .1DQNDGIRQN?SA/OiHER FQSTOt.OG2C TYPM: MO [w7G CiJ06 RM DSPfF.RE7C G1TD3DRiF5 OF D(R1SlitE. Figure ib S. ................................ Q{7NJSE LN GClT. CMpt 1CC OQ15 1iiY1f NllBE]t OF YfMi T1E SPDLSE SPDKM. FOR ALL ST1B.'EL'I5. c=-F-Pf5PMDQJ;'S. NO StRkCL.1= RFS?OK1E?R5: AND 1l7G G'~6 E!'4 DIF?EM34F Gr-970RnS GF DCFCla'...y. x Pa9e m m 70 71 77
Page 12: mmc02a00
m n m a H a a n m 8 E all
Page 13: mmc02a00
: ul H thu th t ~ 0z c i *9 ~ P RO a icu t a m
Page 14: mmc02a00
QD1PIf3t aHE I.TI'DtA2UR€ PStRF1 1 Bt19C~)t~. In 1972, the kepar•t to the Unitad States Gaiqreas on the 'balth uonrerpAnoas or $dklnq for the first time gave serioas oa~sideration to the !ssue of the effects of envirFrs+e-ta1 tobaeeo sxulce on the non- 'rdosr (1). 'lhst others besides ssoke_-s themselvas wald experienoe det--inr^ta1 hialth effects due to tobacro ssoke had been in the aiinis of marty for at least a decabe. Obstetricians in particular scened to have iMCLlifiCd this prlOCClpatiofl, by reportitYQ that cigarette woRinq aevK, pregnant ween .as assoei.ted with prae~turitrf ( 2) and with low bL-rh-wiq)+t (31. Other reports attributed a htele ar_ay of adverse effects of noternal snoklnq an the foetm, ranging from ieasire! ;Y~-na:a1 develcasnt (4,S) to cxqw11•-31 arlfor+ietions (6). lerr studies atter4d ta the effqcts of t wbaccv oa+ooner+ts released into the envirornent at large. Ebr iratar.ce, stvdies in the :ni:ed States (7,e), fhgland (9). Isnel (10) and other ^antries (1:) sraeC !,ow ettilCren of e+dcinq parEnts had Aiarnr Lncifen^e tf S9Q%ZCT~Zoz -i- respiratx. f infecuuns, as well as hiqhw-- hoeoital admissioa rates. laboratory investivatiom sha+ed that expos;ae to meter-nal sroking producwd a defective deyelopient of puLirnaty fuiction in cfiildren, as o,easured by foroed expiratory volume in one seca (12). Finally, reports of childlmd ttasocs aasodated with psr+eetal sa+okinq apQeered in the nediol literatsas. althoisgh otterf with comradic-..cry resulta (13-17). '.riat there were serioas health effects an the adult passive soker - sirh effects as those eaperienaed by the snokers ttemeelves. i.e. lunq oncer - rs not oonsiderad poasible as reoently as a d!mde aqo. This on be gathered from the follovinq paraqrapv published in 1975 s 'In suaeacy, the effaces of cigarette sm~ke on Maalthy rnnrsndotrs aonsist mainly of eiinr eye ard throat irritation. ltarerer, people with certain Nsut ard linq diseases (angina pectcrjs, apppr alloeqie asthom) eay nSfer exwerbation of their syaptaa as a result of oqoatn to totaeeo stdctfilled envirmnnts...' (18). • with resQect to lunq cancer t~heee is no evidence to iidicate whether or not this level of exposure has an effect an the risk of developing lung canoer. Nowever, leouse of the 1or doaaqe and hrief expasara, it wuld seem unlikely that there wuld be a siqnifiont inaease In the risk of dr_-elopinq lunq cencs<: ' 09).
Page 15: mmc02a00
-3- lt is an .stabiished fact that the oKSa11 canaez risrt for nan- mokers is lower than the osnc+er risk of the qe+eral population ( 20- 22). (.'+rnoared to sndcer3, r+a*-smokeeta' risk for cancer is abOUt 501 lcy+l (:C ). Ft•: cYrtvn sites --in particviar the cespiratory tract- rhe differsnes are even eore draaatic. A reoei t etudy, noxtheless, has suqgested that lung cancer asrnq aon-smdcers say be on the rtse. 'ihe suqqescion of such an increase annng non-smdcers has focuse+d i.-rcerest in passive aroking as ane of its possible causes (23). T.us suq,es ion is part/y heseai an a better iadrsstudin9 of the carcinmrnic qualities of Lxoth na:n9tream .nd sidestsasee ciqatette ssoke, as vill be discvss+ed at a later pDint in this chapter. ETutron (23). using data from tro repcesentative saaples of lung cancer .lenths aed sortality statistici at the national leuel, fourld for the period between 1914 and 1968 a 15-told ineaase in lung cancer dtat.1 rates aaonq rhitt rbnsmkin9 .rles ag.d 35-84 ya°rs. 'ftie incease .es the highest for those erles over 65, estinaced as 30- fold. Ebr white femeles the increase has been 7-fold for the 35-94 age 9xc.V• ;trse dwges seee so draastic that the poeiibility of e.rror- in mstrom's est!r.etes has been r.aiseds ;ources go as far back es 191a, and there are ctiecR t.heir re7 iability. 'i'iere are also reasons sone of ais data no rteans by wn!ch to to doubt •le aotvracy of tJ%e ascrrtairnas7t of wokity status in the surveys froe which the data was obtained. Moreover, ttr_ nte,gorization of the smokiry .r3-:s,'Le ckanqes from one source of data to another, as does the ceoaor-.on of .rlf-reported snoking versus sirrc*.late repartinq, thus mirirvq the se of a ocled data .'.:ff1CUi1L tc i.lte:r.rlt. It :,As alaa be-n -.- prapose7d that soAe of the c:unqes described by Fhatscm mey be attributed to better diaqrnees and/or changes in•tumr classificatim. Horever. as the author has pointed out, the trend persists after 1935, when the ea }or changes in d? ~r r i c criteria had been itep 1 ementRd . E•inally, there is the possibility - as Ehstrtm suggests - that the rising rates indicate the itftereention of other fac--=-s in the causation of lung carmoar in additirn to pssonal cigarette saokinq. Other reee.ra+ers do not aqree with fwtram's findiruqs. Garfinkel (24) cwmputed lung c.aneer eortality rates for norr-smolcinq participan•a in tw large prvspective studies. 2n the Merican Canoer Society Study. 9e.000 ean-seokinq sisls and 375.000 norr-sndctnq females were enrolled between OctoEer 1959 atd llarcA 1960. The data used by Garfinioel spw the period between ths ard of the eruvllaent phase and ,7one 30. 1972. '@w secani sv.dy, the Corn Study of Veterans, .as inltiatad in 1954 with the Eailinq of questirnnaires to 293.000 vettrans. Fifty four thousar,as re,porider,rs report.a themselves to be narsaokers. 'lhe foliorvp period endad in 1%9. Mbrtality rates for rm-sokers in bxh of the qrosfs studitl uere estimsted, mainly tluayA the use of death certificates. Fat•es were adjusted to the Urtited States adult ptpulation of 1965. 4rfinkel w u.'+abl e to detect any increase in sortJ i ity due to lung cance.r f or 1?,.ree four-year perioda betreen the years 1960-1972. Ne+aerthe'.ess, he re4,orted a sliqht difference I althiougA not ttattstic.lly significant) in lung cano.r risk for non-..dcl,nq woman married to ssohsrs. as ca:parld to oon-ss+okinq .aars married to ncrr-ssdcers. 'This ss prvbanly the first pooulation study to omsider sucfi an assocci.ation. ;tw results vil l be discvssed in the nrxt sectim. of this ch.pc s-r.
Page 16: mmc02a00
-5- 2 EFIDEMIO[t7CIe '"iUClfS. C~ntrvuersial teports on 1ung canaer trerds anon) norr'saakera have x:t been diiaouraginq erough to dissnisa the nypothesis that passive svoking - also temed involuntary smddng or aecvndarry sraking - iA causally related to lung canaer. In January of 1991, tltt British Mediol Journsl p::blished the first population study s;ecsfica'_1y .ddressi:4 this issue (25). Hisayane r•Q.ott9d tte fidings of a i6 y.e.r old prospectave study which involved a fo11o:+-up fa 91,540 rnrt-=snking s.rned Japanese women. ane hurfCred ard sevc:ty four cases of lung cancer had develooed in this qroF, for which the hust:.nd's s.nking hiet+o:y had been collected ir+depe+dently. wcmen irerzied to heavy soksrs ()-.0 cigrcettas/dsyl snowad a higher risk for lung cancer than .oos+ srrried to rsa-smd:ern (statiardized risk ratio ~ 2.081. t5utheanr.. a statistically aicptiifiont dose-reepanse relationship w found: the relative risk for wives of ex-m.yceta or sroke-s of 1-19 cigarettes/day ..s 1.61 the re/ative risk for heavy 2 s:drars I»201 -+ss 2.0e (Mu:tel-extrnsion X test a 3.299; p, tw tails .30097. The trerd Ms also onservsd .Arn M+sbseds' age ard octtipaticn tias taken into acroslt. TAe highe=t relaive risk. 4.6. ws fourd aaong .omen in agriasltural lasdlies serried to heavy s•dcers aoed 4C-59 years at timQ of ^Arolln:er+t into C:e study. ?U L- _rase in risk for other sa;cr canaers sas obeeried in relation to the h-,uosrris' roking habits. Eti..~ayaa.'s studf reeeiv.+d a great Cesl of atte+tion aeirng borh the medicnl cvmtnity and t4e lav pjbiic. His rest+its ar~d r.et'nds -6-e were closely savt:.ntied. prv:orsng .n eati-anqe of letters either critical or npportive of the study. Mie fact that the Msntel- extensia: test ior ane of the tables .aa ralailated eriono:aly was particvlarly e:phasiud.. A aibseqirnt estinete obtaM+.d by rearrangirig the data .as publiahed supsequently (26). 'R:e new fig:ae d:d not dfsnge the basic cafFlvsion that the study faczcr rd the dis+ease rere signifiantiy asso-+ated. This r••L n.+a., rwxtheless, did not preclude citscis: of the study on different qroamds. EYrst, the research .ork - started in 1966 - did not initially intend to look at the passive rok~lunq cancer asaotiation. Tfus partlcjlar feeture Of the study .ight have affectad the quality of the irlorseticn in several opposite vays. Ch the one h.nd, it sey be that the likeliho©d ol bias in 'repnr-srs3, or 'ir>ut<,ievir,g' was leu than if a well-deuelop.d hypot.hesis me beinq tested. jhis is because the study aubjeets, as well as the intervie.era, rere obviously blird to a hypothesis which wes non-.xistent at the tLm _ of data collection. Cn the other hand. we should ansidez that the resasrch.. not initially interdad to study the effects of pasaive 9"okingo my r+ot have berr: as thocvuqh in the asaer-,a.irmient of exposure to passive snaking as .euld be necessary to detect a nnierate or snril effect. T.tis latter point is not .orrisrn,e, howevc, since it wuld have resultad in a oonservative estimste of the effect. Seco.dly, in Cw origi.•+sl publication by Hirayams no details are given on the way in &nict: deeths .4ere asaertained, nor was there infornation availaCle or: !tie degree of patholaqiwl cv:fiPnetion of We .:ar+ce: cases. It is only tArair,t: sub.equcnt put,liotirns t.'a• we
Page 17: mmc02a00
- I- lezzn that in a sutsen,rle of 23 cases. 17 were adenocarcinoms. Sinoe yd" has been postulated to be aasociated with specific histnloqic typa this infoanet:nn rould seen caieial. Despite -he dra.bncks me'+tioned above. there are certiin features abca.-t Hirayaae s study thst saka it unique and probebiy very difficult to replicate. The prospe_ctiv'e dRsi4n involving a large population in sore tian one geollzaphie reqion is certainly an aeset. Also advsncageous is the fact that the study Was carried out in a traditiaul xtiety vhete ssdcing habits of saen are lov, tlus diminishirq t,he prdb.bilities of includin4 ersmeously clasaified wKn ufio dev.loped cancr_r of the lunq as a eon.+equsnas of their o_n stridry. Al». ia ss>ch a setting earrl.qes my last longer. houses are certainly such snriler. and socialisinq for waen usually takes place irn the osp.ny of cheir hysbsnds. All of the aboue sugqeits that qu.nti.ic++tlon of the hushuds' .okinq habits ary indeed be a good seasure of expos<s•e to seccndhrd rm~lce. Also in 1961. tYichopoulos et. al. published the preliminary results of a study mndu.ted in C.ceeoe (27). t)sing a case eantrol desiqn. trey assemDled a qresup of 40 non-sreking wne+ disgrnead with 'tM cancer in three larye Athens tnepitals. The petients re.:e :.nter•,iewed reaardinq their hunhsnds' ssoRinq habits snd were cavpsred w~th the husCsnds'sidkin9 habits of 149 non-Ardcing hospital controls. 1he oor.trols rer'e ore'apedic petien'-s who did rrt differ siqrufieantly fros the cases with reqard to age. eduation. &ration of a+nr-. i+oe.. occvpecion or resiftncy. "_ft odds ratio for wonen caried -s- to ex-andkrrs ..s 1.8; it .ws 2.4 if they .+ere mnizied to cvzzer:t smdctrs of 1-2J ciqarettes/day; and 3.4 if married to current srokrrs of more than 20 ciqarectes/day. lhis dase-respon.se t d~ fo,ud to be statistically significant (Csi-sq for linear trend - 6.45, p(2 tiils ) t.02. ). 4tw+ exqaassre to hnbsr:is " ssulcinq ou,er the cuyrse of the sarriwge %as assessed a similar txend %.s obxrved. Estiawted odcis ratios were 2.5 for thoes expc+aed to 100 - 299 tjr••-a•+•+ ciqazeEtes ard 3.0 for those eqaoseid to 300 thanard or more cigarettes ( Qii-sq for linear trcrd - 6.50, p(2 tail) t.02).• T.us lincar tzerd ras idential in both qraup of cases: those with citoloqi.cal aslfim.tion of tianr and rld.e with clinical diaqroeis an1y. A later rapart of TridnQoulos' stvdy in 1983 added nearly as m.ny cases, and 509 aze onntrols, to the study sub jects (28). The results .ere suastantially tlae sasrt adds ratio of 2.4 for wives of sedcers of less t1wn 20 cigarettes/day, and odds ratio of 3.4 for wanm s0--ried to seolcen of anre than 20 cl9arettes per day (Chi-sq for linear trend a 6.7, p (2 tail) - .01). The Greeic study, althouqR not entirely orRfarable to that of N.irayame. shom similar results. 2t also shares w+,e of the advantaqes of the latt.er, nau.ly. those ooerCUnirq faaily strueture and fs+euly socz.l life. toqetlrs.r with their iaolications on the assessaen+e of e~osuae. as err.n.rated abow. Differer.t elthaioloqical problens., r~a.ever. ought to be oonsidered. Fbr instanoe, the study qrvup is saeLl and the dcqrx of patholoqiol ooifizmstion poor (only 651 fad a::toloQial or hir.olopcal di.gnais). The study suojects (norr-
Page 18: mmc02a00
-9°q suoairyq wmen vit.1 lung cancerl reoresant aoout 781 of all the female lung carcer cases diagroeed in t.he pnrLicipstinq hospitals. As is ,rencioned i,n a subseqirnt section of uLis ctapter isee: Histoioqic ^.yx eni l.ung Cancer) the propor:ion of rnrtsmdcu+q parients in other series of la+.) cancer eases has bern'+stst+lished at aboit 10% for .aecn. '1!he exwsively high proportion reported by T:ichopou!aa suqgests that saee of his cases ray vc.'eei have been saokers vlio did .wt a3ni' to theic tahscco habits for social or other reasons. or, alttu,yh tne proFanition is soue.Aat unlikely. it msy be that non- sndc..-x7 .m-en with lung canoar in Greece are eore likely to be hosoitr:iud than in other parts of the world. In addiLion, the fact t.hat both cases and arnrols wsre sterfie.4d by a sinqle, unblirded eYsaarelt'er. points tovard the possibility of misclassification of eaposu.^e. Cn the otlver hand. the attertot to assess exQosure according to spouses' ssdtinq habits during the total length of nerried life ney have provided mozz ar•p~ete ascertnimrnt of expouse. Pelayo Corr.a at. al.. using + ease-oontrol design, assessed the risk of lung cancer for both non-stnokinq s+en a,id wwn with regard to the s,ok.inq habits of tl+eir smLses artl parents (29). '1Tey tound a.i odds ratio of 2.0 for e.en svuried to uv-.Ten who snoked naxe than one pecl: o' .iyarettes a year. Ztw opcrpporLirg adds ratio for 22 wmn -10- of of iusy canoer, as indimted by an odds ratto of °"1.a1 (Opocsite resulis on this parental amkinq-lunq cancer assorition have been reported ny Sandler, et. ai.. Thoee results vili be disauaed latter in this ctapte.r).. Br .ffect observed. Ao.e,+er. .as neither siqnificant nrs seensd to be doarrelat.d. 1ten exposuae to both parents` snokinq .es oabined, the o3ds ratio becane siqnificm only for tJrse cases .•xpc»ed to - > 41 paclcs of cigarettes per year (CR - ).11, p < .05). Inportant dra.,e.cls can be !ourd ia the study by Pelayo Car-es et. al.. Its sr.in disadvantage is the faet ttat the r.aber of partucipents is so swell that any of the pnasible stre qt.1s of the study - sueh as the 100% histoloqie eonfira.tion of fail to conpernata for this we.)m.ss. d!agrxetY. - A aore recen t epidenioloqie study on the lung cancer-passive smoking relationship r.s car-ried out by CarfiAkel. Auerbsch and Joubert (30). Their qroup of cases w ecaVrisad of all fenale non- md" lung cancer patients diaqno.ed in three hospitals in Nr.i Jersey and oFle hospital in Ohio. laaien diagrnsnd with colon cancer se.^ved as cant.rols. ".Tree oontrols rere satet~ed to eacfi case on the basis of age (-5 years) ard hospital of diaqrnais. Patholoqy slides for both cases and xnerols .ete reviewed blirdly in order to cmfirm the diagnosis. Cne hiaxlred ard thirty lour lung canaer cases and a02 Tti:•ed to s+rok:nq aen •as 2.07. tlrithar .as statistically oontrols were then uttem-e..ed reqartilnq the aedcinq habits of their siqrifiepnt. After oon_rolliAq far the spou.ses` smdcinG ha5it., husbands. Clurent s.mkinq habits. rxsk~er of cigarettes sadxd at iroce-nal (but rot paternal) s7dcinq was assoaattd %n.th a hta:t- r:--k ha+e. and m.noer of vQars tne husbrnds had snoked .ere assessed. E:'• 1' • At ~. 4t t
Page 19: mmc02a00
-11- Sutilarly, wm+ wele questaoned abait tne numrs of hours a day they had oeen ezpased to the sndke of others in the psst S and 25 years. Lt w%s found that wnen merriad to saokers of :0 or nure c:garettes prr day h.d an odds ratio of 1.99 (95% CL:1.13-3.50) as carpazed to tlvi+e married to eom-mokers. If half that aaount of ciqirrttes .ea sad:ed at hone, the o31s ratio -.as estiswted to be 2.11 IySSC.:1.13-3.951. Similarly# an its3eased risk wes fourd for thase .orrr nwrried to een rho had ssdced for 20-29 years (oR - 2.2). but, t mr~_maRedly, not for tlnse ewrried to ssolo.rs of sore than 30 years. No ixrease of risk +rs foisd to be deFd+de+t an the ra>ffter of e,poe:ce-Aours in the last five and 25 years. (sinq logistic regression txtniquea., a model vas fitted to the stsiy data. 3'tw sndel included te=w+ for posaible oa+foLaiders "sev-ral oontirnious variables represenEii+g e:@oa:re. An inat+ase in risK with increasing exposure was fotis:d foc the nuiber of cigarettes smoked by the txtsbar+d at hant. 'lhis trerd .as statistically siqni.`icant at the 0.0S level. 11,e study by Garfinkel et. al. atterpted to tackle sa++e of t.`ie metlodoloqical problene present in t1e fors+er stndies on passive snoL,inq. ".tie aurl+ors were successful in puttinq together a siu+ble ,_arp cf ases .tio w.re cvLir:Kd never-anoiters and whose disqncsis ws tLatologially co.firtrod. 'ltair asArssnent o[ exposure ~.as nct 1 uiitod to the a~e derived fran the huabe.zis ' somlctno but included :.'at experierx.zd at " worksite and cr_her ateaa. vok+er'.ul s:atisucal tec!^niqnes (loyistic regression) were used to ana!}ze the -UtA, ordLcuvq results c^v:scen.t with L'x fltiv+as of t•,e cw :::,r: o~,ezTsCzo% i+r ..nIL -.12- ma7or atuCies on this toQic. It sews. itr-_refore, that the only aspect of the stvdy that oo•+ld have af:ected the results is that 6tuch is nost difficult to evalustet the reliahiLity of the data cotlected thrurqh the interviev. Eyan thouqh t!r itterviwiers .are n.qoasedly blind in raqard to tha hypatheziA ot the stvdy and the case/czritrol statxu of the paztieip.nta. the pactidpanes tfryaselves .ce .rara of theu om cmnditSan. In this situation tMe poasihiiity of overzeporting by the aw .ust be bQne in aunG. Additionally interNlers of surrogate respandeflt.7 ary heV" atflCt ld the reliability of the information about the exposure variable. in fact. Msntel (31) has shawn that Crrtinlo.l 'a es•-iaatar of the odds ratioa vary ooaiderably +caordlr9 to the tyPa of zst.pondent. Be found t.hat the adis rat.io. for lua+q ciwcer u.s 2.S7 (95% a.: 0.6t-15.2t) in the subqroW in .hich the respordertt,. vere eithtr a drr7ther or a sm of the st.udy sub jects. In the wibgroup in .ftich the lnulrni provided inforaistion on the exposure he fou.d thst the adds ratio as o.77 (95sQ.: O.S6-.1.06. Atd finally, in the subqtoup of cases who reported about their own eapauae. the odds ratio .as eatSated to be 0.81 (95% Q.: 0.56-1.3e). At the ssne tae tlat this rtiport .as being prepsred, tw additional epideeioloqic studies claiauq the euivtence of a sodrsste effect of passive aeokinq rn l:nq eaneer risks wre publisAed. Tfie first of these w also eorkttted in Japan (32). It is a cas-e- crntrol study. bat, as w true of its prs,deoeser. it nekes use of a ed,ort :.littally defined for anotf.er piupoue (the study of the effects of a;cmic rsdiation in 41_-oertuu and Xirpisaw-i sur.•!vcss) in orL'e- _o
Page 20: mmc02a00
ixiv. -1]- obtain ics study sub jec_s. 4he cohst. crn4rised of ahout 110.000 renpla, yielded 525 naes of lunq i:ancrs between 1971 and 1980. pacholoqic dieqr+osis weie dcne in 57% of these. Controls .se selr_ted fstm tne r.ane cohort amem9 memoers `'ho had ^ot develop l:ng cancet. TfiE controls `+ere individually ertd:ed to the cases with regard to year of birrh (- 2 )t•s), casidenoe. sex, vital status. ard wr,etr-s or not t.h,y had belonged to the conor' subqro:p that und..rtient Siernial me.diral exaednstions. tnfornrtion res abtained for 428 cases and 957 controls. Ho+ever. only 19 aiale osao ar+d 94 fee.le cases ocruczed .nocq ronyrok,ess. Mer+g the ron-smok.inq nales tlre odds ratio associated with hsvinq a spoiue ++fo sioiaed .as 1.8 (90s Q.: 0.5 - 5.61. Ebr feseles the mriespondin4 figure ras 1.5 1901 Q.: 1.0 - 2.S). in adlition. [eTales show.d an itrrasse in risk of lunq canocr with the inc'ensinq tenber of cigarlttes sn6ksd by the MtsEend: the odds ratio oeing greater than 2.0 for the higMet expoaue category. 7tw test of linear trend. ho+eMer• vs not statistically significant at the 0.05 level. Likewise. araother eies.sure of Mnesnds' snflk.inq -- the nurber of years the husband miehs! - w rot statistically siqnifiant. rirrlly, their results shaed that non-ssdung wbn ++ho had been eiqoeed to the huelmsts' asakin9 in the last 10 years had a laer risk lCR  1.3, 901 Q.: 0.9 - 2.4) than those w:+n eyq+ased within the last 10 years (OR • 1.11, 9% Q.: 1.0 - 3.2). Sa-ie of the 3bove estin.trs are consistent in their meqnitude 'if not in _heir r..atistimi siqnificancel with the results of the ':rst Jap3r*se study, as .el: as with the ot`er studies rn pessive y.-. n x,:ny. ~C..eve_, :his subse.velt sn:oy equally shares in some of the Z4SZTSEzO% -14- eetlndologic pcobleae that cast daubts over•tAeir results. Cne ^~ pnobleo is the poor rste of histoloqic oonfirsetion. 1lie authors, in trying to uKumve+t this prablm. mrried onc seoarnte anelysis for those vith patholoqie diag:osis, as we11 as thase vithwt it. An urd-ired aormecltxnoe of sud: apptoeel:, ha+ever, is tlhe losa of statistical poi+er due to the reaultltq smaller qroips available for es*;nation of the psrraters. Mother potertetatly impoEtartt problea: ca+oerns the obtaininq of the info~tion atart exposiae. Only 22 cases ard 26 ca:tsols out of the t,otal, altteuqh rvt neoaaarily out of the non-s.okinq 9roup only. were avaaable to be irrtervietie,d. Dwsure far all others w as.sensed throuqh the intervie.rinq of Aw=°4au r'e3Fu^des +ti. Even tharlh the distribut.ion of types of siaz".x)ete respendv+ts w.iniiar for oses and evnrols, it cannot be ensured that selective recall .s elimirrttl. A final point omrns the pecvlarities of the study qro:p. snavivors of the ata:dc erplosions of Bireshiae and M.qasaki. Sinoe both cases ard oontrols were equal in that reqard. we do not quest.ion the internal validity of the findings Eeycnd the .ethadolaqic probleas sentioned. No.ever. we should be cautioue in any attmpt to extrapolate these results to the ~ general powlation. It oould be that the effect ctaerved is only a pccduct of a synergistic effect bet.ean rad.iatioa and oqosure to tob-aco sedoe. ^!1is synetyy, of ctiurse, vould not be possible in the yer+eral pcpulation sii+ee oe the vhole it lacks eaposure to the levels of radiation emyaerienasi by the subjects in this study. Finally, the autlxus of this stu&/ have been less then o3»ervstiue in astAblisRirq that level of statistical siqnificwne whidh tlry are villing to aocept. Yalues of p beb+en 0.05 a.~d 0.1 are reported as

Text Control

Highlight Text:

OCR Text Alignment:

Image Control

Image Rotation:

Image Size: